Table Of ContentANNALS OF CARNEGIE MUSEUM
VoL. 65, Number 1, Pp. 27-88 22 February 1996
A PHYLOGENETIC REASSESSMENT OF THE TADORNINE-ANATINE
DIVERGENCE (AVES: ANSERIFORMES: ANATIDAE)
Bradley C. Livezey
Associate Curator, Section ofBirds
Abstract
Phylogenetic analysis of eight problematic genera of waterfowl {Stictonetta Plectropterus, Sarki-
diornis, Hymenolaimus, Tachyeres, Merganetta, Malacorhynchus, and Salvadorina) and representa-
tives ofother tadornine genera and anatine tribes was undertaken using 114 morphological characters
(39 skeletal, four tracheal, 11 natal, and 60 of the definitive integument) and a variety of weighting
schemes, in anattempttoclarifyrelationshipsin themostpoorlyunderstoodsegmentofthephylogeny
ofAnseriformes(Livezey, 1986<3). Underequal weightingofcharacters,40 equally parsimonioustrees
were found that shared the following findings: (1) a sister-group relationship ofStictonetta to a clade
comprising the Tadorninae and Anatinae; (2) a sister-relationship between Plectropterus and Sarki-
diomis; (3) monophyly of the typical shelducks and sheldgeese (Tadorna, Cyanochen, Alopochen,
Neochen, and Chloephaga); (4) monophyly ofHymenolaimus, Tachyeres, and Merganetta-, and (5) a
sister-relationshipbetween Salvadorina and Malacorhynchus. An analysis wherein skeletal characters
were given twice the weightofothercharactergroups resultedin fourshortesttrees which, in addition
to the findings under equal weighting, indicated that: (1) Plectropterus + Sarkidiornis represent the
sister-group ofthe typical shelducks and sheldgeese; (2) a sister-relationship between the clade com-
prising Hymenolaimus, Tachyeres, and Merganetta and the preceding clade, collectively constituting
the “tadornine” clade; and (3) a sister-relationship between Malacorhynchus P Salvadorina and an
“anatine” clade comprising the remaining tribes and subtribes (Cairineae, Nettapodeae, Anateae, Ay-
thyini, Mergini, and Oxyurini). An analysis in which skeletal, tracheal, natal, and definitive characters
were assigned weights of four, three, two, and one, respectively, produced identical results to the
preceding weighting scheme. Successive weighting by rescaled consistency indices produced two
shortest trees that resembled those found under the two preceding protocols except: (1) the Mergini
wereplacedas the sister-groupofthe “tadornine” cladedefinedabove; and(2) theCairineae(Cairina,
Pteronetta, and Aix) were placed as the sister-group to Plectropterus + Sarkidiornis. Based on the
comparative strengths ofhypothesized homologies ofskeletal characters and assessments oftopolog-
ical support(includingbootstrapping,decay indices,andnear-shortesttrees)forall weighting schemes,
aconservativeclassificationwas basedon theaprioriweighting schemes favoring skeletal characters.
Maps of selected ecomorphological attributes on this phylogenetic tree revealed several distinct evo-
lutionary trends, but incompletely resolved relationships among these diverse genera and within su-
pragenerictaxalimitedinferences. Additionalpromising areasforanatomicallybased systematic study
are identified, and the need for continued cladistic study of this poorly resolved but topologically
critical segment ofthe phylogeny ofAnseriformes is emphasized.
Introduction
Greater diversity of opinion has attended the classification of the following
eight southern-hemisphere genera of waterfowl than in any assemblage of similar
size in the Anseriformes: Stictonetta, Plectropterus, Sarkidiornis, Hymenolaimus,
Tachyeres, Merganetta, Malacorhynchus, and Salvadorina (Table 1). Early tax-
onomists typically included these genera within larger, more-familiar groups of
waterfowl on the basis of superficial comparisons and very limited material
(Eyton, 1838; Gray, 1841, 1871; Bonaparte, 1856; Sclater, 1880; Salvador!, 1895;
Phillips 1922, 1923, 1925, 1926).
In recent decades, taxonomic classifications of the Freckled Duck {Stictonetta
Submitted 16 June 1995.
27
—
28 Annals of Carnegie Museum VOL. 65
Table 1. A century of taxonomic classifications of eight taxonomically problematic genera of
Anatidae.
Genus Salvadori(1895) Phillips(1922-1925) Peters(1931)
Stictonetta Anatinae, between Mar- Anatinae"*, between Anas Anatinae, between Casar-
maronetta and Hetero- and Heteronetta. ca and Tadorna.
netta.
Plectropterus Plectropterinae, before Plectropterinae, before Plectropterinae, withAn-
Cairina and Sarkidior- Cairina and Sarkidior- seranas.
nis. nis.
Sarkidiornis Plectropterinae, between Plectropterinae, between Anatinae, between
taxa currently merged taxa currently merged Neochen and Cairina.
into Cairina. into Cairina.
Hymenolaimus Merganettinae, between Merganettinae, between Nyrocinae% between Po-
Salvadorina and Mer- Salvadorina and Mer- lysticta and Salvadori-
ganetta. ganetta. na.
Tachyeres Fuligulinae"', between Ay- Fuligulinae"*, between Ay- Nyrocinae, between Ay-
thya and Bucephala. thya and Bucephala. thya and Bucephala.
Merganetta Merganettinae, with Sal- Merganettinae, with Sal- Merganettinae (monotyp-
vadorina and Hymeno- vadorina and Hymeno- ic), after “Merginae”
laimus. laimus. (mergansers).
Malacorhynchus Anatinae^, between Anas Anatinae"*, between Anas Anatinae, between Anas
(part) and Marmaro- (part) and Aix. (part) and Rhodonessa.
netta.
Salvadorina Merganettinae, with Hy- Merganettinae, with Hy- Nyrocinae, with Polystic-
menolaimus and Mer- menolaimus and Mer- ta and Hymenolaimus.
ganetta. ganetta.
^Generic taxa used by author(s) are replaced by currently accepted synonyms.
Johnsgard (1979) listed no tribes within subfamilies.
Salvadorina merged withAnas by author.
^Skeletal specimens ofSalvadorina not examined by Woolfenden (1961).
naevosa) have included placements as: an aberrant dabbling duck (Delacour, 1938;
Delacour and Mayr, 1945) or shelduck (Boetticher, 1942, 1952); the sole member
of a tribe or higher taxon closely related to the true geese (Verheyen, 1953, 1955;
Frith, 1964<3, 1964^?, 1965, 1967; Johnsgard, 1978); a very primitive sister-group
of the majority of modern ducks, geese, and swans (Frith, 1967; Olson and Fed-
uccia, 1980; Madsen et aL, 1988; Sibley and Ahlquist, 1990; Sibley and Monroe,
1990); or as both very primitive and a close relative of the stiff-tailed ducks
(Fullagar et ah, 1990; Marchant and Higgins, 1990). Also, assignments of single
species to monogeneric higher taxa has led to the phylogenetically misleading
conclusion that each has, like Stictonetta, “no near living relatives” (e.g., Johns-
gard, 1978:xx) or “is not closely related to any other tribe” (e.g., Edkins and
Hansen, 1972:105). However, under an assumption of the monophyly of the biota
of the earth, any species is the sister-group (i.e., closest relative) of some other
group of organisms. At present, there is a consensus among most systematists
that Stictonetta should be excluded from the true ducks (Anatinae), although other
details of its relationships within the Anseriformes remain in dispute (Wolters,
1976; Johnsgard, 1979; Livezey, 1986^?; Madsen et ah, 1988; Sibley and Monroe,
1990).
The Spur-winged Goose (Plectropterus gambensis) and Comb Duck (Sarki-
diornis melanotus), traditionally considered to be members of an assemblage
termed the “perching ducks,” or formally the Subfamily Cairininae or Tribe Cair-
inini (e.g., Delacour, 1938, 1959, 1964; Delacour and Mayr, 1945; Boetticher,
—
1996 LiVEZEY TADORNINE-AnATINE DIVERGENCE 29
Table 1. Extended.
DelacourandMayr(1945) Boetticher(1952) Delacour(1954—1959)
Anatinae, Anatini, as “aber- Anatinae, Tadornini, in mono- Anatinae, Anatini, after Hymen-
rant.” typic “section.” olaimus, and Merganetta.
Anatinae, Cairinini, after Sarki- Anatinae, Cairinini, before Anatinae, Cairinini, after Sarki-
diornis and Cairina. Cairina and Sarkidiornis. diornis and Cairina.
Anatinae, Cairinini, between Anatinae, Cairinini, between Anatinae, Cairinini, between
Nettapus and Cairina. Cairina and Nettapus. Nettapus and Cairina.
Anatinae, Anatini, as “aber- Anatinae, anatini, afterMala- Anatinae, Anatini, between Ma-
rant.” corhynchus. lacorhynchus and Merganetta
Anatinae, Tadornini, as “aber- Anatinae, Tachyerini, monoge- Anatinae, Tadornini, afterLo-
rant.” neric. phonetta.
Anatinae, Merganettini (mono- Anatinae, Oxyurini, after Heter- Anatinae, Anatini, between Hy-
typic), after Oxyurini. onetta. menolaimus and Stictonetta.
Anatinae, Anatini, as “aber- Anatinae, Anatini, before Hy- Anatinae, Anatini, between
rant,” menolaimus. Rhodonessa and Hymenolai-
mus.
Anatinae, Anatini, merged with Anatinae, Anatini, between taxa Anatinae, Anatini, merged with
Anas. currently merged into Anas. Anas.
1952; Johnsgard, 1965<2, 1978), were reassigned to a monotypic subfamily (Plec-
tropterinae) and the shelducks (Tadorninae), respectively, by Livezey (1986a).
Similarly, steamer-ducks (Tachyeres spp.; Livezey and Humphrey, 1992) and the
Torrent Duck (Merganetta armata), for which persistent doubts concerning phy-
logenetic relationships were finessed by some through allocations to monotypic
tribes (Delacour and Mayr, 1945; Verheyen, 1953; Johnsgard, 1965a, 1966, 1978),
also were included in the poorly resolved Tadorninae by Livezey (1986a).
The Blue Duck (Hymenolaimus malacorhynchos) and the Pink-eared Duck
(Malacorhynchus membranaceous), traditionally considered to be aberrant mem-
bers ofthe dabbling ducks (Delacour, 1938; Delacour and Mayr, 1945; Boetticher,
1952; Delacour, 1956; Johnsgard, 1965a, 1965Z?, 1978, 1979; Kear and Steel,
1971; Sibley and Monroe, 1990), were included as well among the tadornines by
Livezey (1986a). However, Olson and Feduccia (1980:22) suggested that Mala-
corhynchus, together with Stictonetta and perhaps Merganetta, were '‘primitive
relicts of an early anatid radiation and may be only distantly related to the re-
mainder ofthe Anatidae.” Salvadori’s Duck (Salvadorina waigiuensis), for which
skeletal and tracheal specimens were not available in the analysis performed by
Livezey (1986a), also has been classified in recent decades as an unusual member
of the dabbling ducks, and often merged with the genus Anas (Mayr, 1931; De-
lacour and Mayr, 1945; Delacour, 1956; Johnsgard, 1965a, 1978, 1979). Subse-
quent behavioral comparisons (Kear, 1975) and limited osteological data (MIL
kovsky, 1989), however, indicate that the species may be more closely related to
other waterfowl variably specialized for lotic habitats {Merganetta and Hymeno-
laimus) or to the unique Malacorhynchus. Based in large part on these data,
Livezey (1991) tentatively included the genus among the shelducks and allies
—
30 Annals of Carnegie Museum VOL. 65
Table 1. Continued.
Genus Woolfenden(1961) Johnsgard(1965“) Wolters(1976)
Stictonetta Anserinae, Dendrocygnini Anserinae, Stictonettini Stictonettinae (monotyp-
(not examined). (monotypic), after An- ic), between Dendro-
serini. cygninae and Anseri-
nae.
Plectropterus Anatinae, Tadornini, be- Anatinae, Cairinini, be- Plectropterinae (monotyp-
fore Chloephaga and fore Cairina and Sarki- ic), between Tadorni-
Cyanochen. diornis. nae and Anatinae.
Sarkidiornis Anatinae, Anatini, before Anatinae, Cairinini, be- Anatinae, before Cairina
Pteronetta and Cairina. tween Cairina and and Pteronetta.
Pteronetta.
Hymenolaimus Anatinae, Anatini, be- Anatinae, Anatini, before Anatinae, between Mala-
tween Malacorhynchus Merganetta and Anas. corhynchus and Tachy-
and Tachyeres. eres.
Tachyeres Anatinae, Anatini, after Anatinae, Tachyerini, be- Anatinae, between Hy-
Malacorhynchus and tween Tadornini and menolaimus and Mar-
Hymenolaimus. Cairinini. maronetta.
Merganetta Anatinae, Merganettini, Anatinae, Anatini, be- Anatinae, between Lo-
monotypic. tween Hymenolaimus phonetta and Salvador-
and Anas. ina.
Malacorhynchus Anatinae, Anatini, be- Anatinae, Anatini, be- Anatinae, between Salva-
tween Salvadorina and tween Anas and Mar- dorina and Hymenolai-
Hymenolaimus. maronetta. mus.
Salvadorina Anatinae‘*, Anatini, be- Anatinae, Anatini, as Anatinae, between Mer-
tween Anas and Mala- monotypic subgenus of ganetta and Malaco-
corhynchus. Anas. rhynchus.
(Tadorninae). Thirteen representative classifications ofthese eight genera spanning
the last century are summarized in Table 1.
Woolfenden (1961) presented the first study of the comparative osteology of
waterfowl from the perspective of(phenetic) assessments ofrelationships. Livezey
(19866) presented a species-level phylogenetic analysis of Tachyeres based on
morphological comparisons, one which compared favorably with an assessment
of genetic distances in the group (Corbin et ah, 1988) and biochemical patterns
in secretions from the uropygial glands (Livezey et ah, 1986). Other contributions
to the knowledge of the anatomy of these genera include, by genus: Stictonetta
(Campbell, 1899; Frith, 19646; Olson and Feduccia, 1980), Plectropterus (Rand,
1954; Moller, 1969), Sarkidiornis (Garrod, 1875), Hymenolaimus (Kear and Bur-
ton, 1971), Tachyeres (Livezey and Humphrey, 1986, 1992), Merganetta (Nie-
thammer, 1952; Weller, 1968; Raikow, 1970), Malacorhynchus (Frith, 1955,
1967), and Salvadorina (Rothschild, 1895; Mayr, 1931; Kear, 1975; Mlikovsky,
1989). Tyler (1964) included several of these genera in a comparison of the mi-
croanatomy of anseriform egg shells. Fossils assignable to any of these eight
genera are limited to Pleistocene material of Sarkidiornis and subfossil material
ofMalacorhynchus (Brodkorb, 1964; Howard, 1964a; Olson, 1977, 1985; Worthy,
1995).
—
—
1996 Livezey Tadornine-Anatine Divergence 31
Table 1. Extended.
Johnsgard(1978) Johnsgard(1979)*’ Livezey(1986") SibleyandMonroe(1990)
Anserinae, Stictonettini Anserinae, between Stictonettinae (mono- Stictonettinae (mono-
(monotypic), after Cereopsis and Tador- typic), between An- typic), between Ox-
Cereopsini (monotyp- nieae. serinae and Plectrop- yurinae and Cygni-
ic). terinae. nae.
Anatinae, Cairinini, be- Anatinae, before Cairi- Plectropterinae (mono- Anatinae, Anserini, be-
fore Cairina and Sar- na and Pteronetta. typic), between Stic- tween tachyeres and
kidiornis. tonettinae and Tador- Cairina.
ninae.
Anatinae, Cairinini, be- Anatinae, between Tadorninae, Sarkidior- Anatinae, Anserini, be-
tween Cairina and Pteronetta and Net- nithini (taxon tween Pteronetta and
Pteronetta. tapus. amended; monotyp- Nettapus.
ic).
Anatinae, Anatini, be- Anatinae, between Tadorninae, Tadornini, Anatinae, Anatini, be-
fore Salvadorina^. Merganetta and Sal- in subtribe with tween Merganetta
vadorina'^. Merganetta and and Salvadorina
Tachyeres.
Anatinae, Tachyerini, Tadorninae, after Ta- Tadorninae, Tadornini, Anatinae, Anserini, be-
between Tadornini dorna as last includ- in subtribe with Hy- tween Tadorna and
and Cairinini. ed genus. menolaimus and Plectropterus.
Merganetta.
Anatinae, Merganettini, Anatinae, between Tadorninae, Tadornini, Anatinae, Anatini, be-
between Cairinini and Amazonetta and Hy- in subtribe with Hy- tween Amazonetta
Anatini. menolaimus. menolaimus and and Hymenolaimus.
Tachyeres.
Anatinae, Anatini, be- Anatinae, between Tadorninae, Tadornini, Anatinae, Anatini, be-
tweenAnas and Mar- Anas and Marmaro- as subtribe between tweenAnas and
maronetta. netta. Tadorna and sheld- Marmaronetta.
geese.
Anatinae, Anatini, after Anatinae, after Hymen- Anatinae‘*, Anatini, pro Anatinae, Anatini, be-
Hymenolaimus as olaimus as first visionally included tween Hymenolaimus
first member ofAnas. member ofAnas. in Anas. andAnas.
Phylogenetic analyses using molecular characters are considered by some to
hold potential for especially reliable inferences of phylogeny, but an assumption
of inherent superiority lacks empirical support (Hillis, 1987; Sanderson and Don-
oghue, 1989; Moritz and Hillis, 1990). Moreover, comparisons of studies based
on molecular and morphological data, even where fundamentally phenetic meth-
odologies were applied, share broad areas of agreement (e.g., Bledsoe and Rai-
kow, 1990; Omland, 1994; see also Swofford, 1991). Unfortunately, most of the
larger studies of Anseriformes employing molecular techniques included none of
the eight genera considered here (Kessler and Avise, 1984; Patton and Avise,
1985; Scherer and Sontag, 1986). Numachi et al. (1983) included but a single
member ofthis group ofgenera, Sarkidiornis, precluding any tests ofrelationships
among members. Bottjer (1983) included Plectropterus and Sarkidiornis in his
immunological comparisons; one especially distinctive finding was the low sim-
ilarity between Plectropterus and the “perching ducks” Cairina, Chenonetta, and
Aix. The electrophoretic comparisons of egg-white proteins by Sibley and Ahl-
quist (1972:fig. 10-12) included Plectropterus, Sarkidiornis, Hymenolaimus, and
Malacorhynchus, but not even a phenetic analysis of the intriguing patterns was
presented. Other molecular analyses including two or more of these taxa also
were phenetic in methodology, and therefore any inferred associations represent
32 Annals of Carnegie Museum VOL. 65
assessments of overall similarity rather than direct indications of phylogenetic
relationship (Jacob and Glasser, 1975; Brush, 1976; Jacob, 1982; Jacob and Hoer-
schelmann, 1993). Of the latter, the comparisons of feather proteins by Brush
(1976) included six of the eight genera of interest here, but the resultant banding
patterns proved difficult to interpret phylogenetically and rendered tribal assign-
ments problematic for Stictonetta, Hymenolaimus, Merganetta, and Malacorhyn-
chus.
The analyses by Sibley and Ahlquist (1990) using DNA hybridization, and the
related classification (Sibley et ah, 1988; Sibley and Monroe, 1990), also lacked
any samples from these eight genera. The placement of Stictonetta by Madsen et
DNA
al. (1988) in an analysis of 13 species of Anseriforrnes using hybridization
must be viewed with caution in light of criticisms of this general methodology
(Cracraft, 1987; Houde, 1987; Sarich et ah, 1989; Springer and Krajewski, 1989),
poor representation of taxa, and unfavorable metric properties evident in the com-
parisons.
The primarily osteological study by Livezey (1986<3) remains the only work in
which even a majority of these problematic “tadornine” genera were analyzed.
However, support for inferences in the topological vicinity of the divergence of
the “shelducks” (Tadorninae) and the “surface-feeding ducks” (Anatinae) in the
preliminary analysis by Livezey (1986<2), should be considered tentative. This
assessment stems from several concerns: (1) the polychotomous topology depicted
for the genera included in the tadornines (Livezey, 1986<3:fig. 1, 3); (2) the limited
numbers of characters supporting groupings within this segment of the tree, e.g.,
those uniting Cyanochen with Hymenolaimus, Merganetta, and Tachyeres; (3) the
impossibility of an assessment of support (e.g., bootstrapping of characters or
consideration of suboptimal topologies) or the treatment ofpolymorphic character
states at the time of the analysis;—(4) three characters used by—Livezey (1986a) to
support clades in the tadornines characters 15, 33, and 61^ were judged unre-
liable for these taxa based upon further study (Livezey, unpublished data); (5)
several of the characters of the pelvic limb (including characters 56 and 69) show
significant homoplasy evidently associated with diving habit (Livezey, 1986a;
Faith, 1989), therefore groups united in part by these characters (e.g., Hymeno-
laimus, Tachyeres, and Merganetta) merit confirmation; and (6) the profound
limitations of software and computing platform increased the likelihood of sub-
optimal solution sets (e.g., retention of more than 50 trees in memory for this
data set typically caused a “crash” of the analysis).
Livezey (1986a) indicated several areas ofpoor resolution and marginal support
in the resultant phylogenetic hypothesis, and subsequent study discovered alter-
native solutions to at least the basal segment of the tree (Livezey, 1989). In
addition, reanalysis ofthe original data matrix compiled by Livezey (1986a) using
improved versions of analytical software discovered multiple, slightly shorter al-
ternative trees including several distinct topologies (J. Harshman, personal com-
munication), the majority ofwhich were the result ofpoorly resolved relationships
in the vicinity of the divergence of “tadornine” and “anatine” genera. With
respect to the eight genera of concern here, these alternative trees shared the
following elements: (1) paraphyly ofStictonetta and Plectropterus to “tadornine”
and “anatine” genera (as inferred by Livezey, 1986a); and (2) placement of Ma-
lacorhynchus. Blue-winged Sheldgoose {Cyanochen cyanopterus), true shelducks
(Tadorna) and other sheldgeese (Alopochen, Neochen, and Chloephaga), Sarki-
—
—
1996 Livezey Tadornine-Anatine Divergence 33
diornis, and the clade comprising Hymenolaimus, Merganetta, and Tachyeres as
paraphyletic to the genera included in the Anatinae by Livezey (1986^).
Contradictory assessments of relationships in this group of genera “transition-
al” between the anseriform subfamilies Tadorninae and Anatinae, as well as short-
comings in analyses to date, justify a renewed effort to recover phylogenetic
structure in this important group of waterfowl. Because of the phylogenetically
central position ofthese genera, i.e., their topological proximity to the divergences
ofthe tribes ofthe Anatinae (Livezey, 1986r?, 1991), a study ofthis kind provides
an opportunity to assess available morphological evidence concerning the rela-
tionships among the anatine tribes as well. This paper, one of a series of finer-
scale analyses of the major subparts of the phytogeny of waterfowl (Livezey,
1991, 1995a, 1995A 1995c, in press a, in press A in press c), presents a phylo-
genetic analysis of the eight aforementioned genera and representatives of related
tribes, using an enlarged set of morphological characters and improved analytical
algorithms. This analysis is followed by ancillary considerations ofmorphological
evolution, phylogenetic patterns of behavior and ecology, implications for histor-
ical biogeography, and promising areas for future study for this most challenging
aspect of the phytogeny of Anseriformes.
Materials and Methods
Taxonomy
—
Composition of the Ingroup. Determination of the broad limits of the com-
paratively derived members of the Anseriformes, based on the cladistic analyses
by Livezey (1986a, 1995a, in press c), conservatively included representatives of
all groups exclusive ofthe screamers (Anhimidae), Magpie Goose (Anseranatidae:
Anseranas semipalmata), whistling ducks and allies (Dendrocygninae), and true
geese and swans (Anserinae). Comparisons included representatives of all taxa
included in the eight genera of primary interest Stictonetta, Plectropterus, Sar~
kidiornis, Hymenolaimus, Tachyeres, Merganetta, Malacorhynchus, and Salva-
dorina. Representatives of other “tadornine” genera (Cyanochen, Alopochen,
Neochen, Chloephaga, and Tadorna), also were included, which were inferred to
constitute a monophyletic group in this series of phylogenetic analyses (Livezey,
in press a). The Cape Barren Goose (Cereopsis novaehollandiae), considered by
some to be an aberrant sheldgoose or an evolutionary “link” between the true
geese and the sheldgeese (Delacour and Mayr, 1945; Delacour, 1954; Johnsgard,
1961a, 1978; Woolfenden, 1961; Frith, 1967; Rear and Murton, 1973), is excluded
from this study on the basis of its inclusion among the true geese and swans
(Livezey, 1986a, 1989, in press c).
Also included are “basal” representatives (see below for methods of charac-
terization) of the major clades of Anatinae: three subtribes of dabbling ducks or
Anatini (Cairineae, Nettapodeae, and Anateae; Livezey, 1991), as well as the
pochards (Aythyini; Livezey, in press b), seaducks (Mergini; Livezey, 1995^),
and stiff-tailed ducks (Oxyurini; Livezey, 1995c). Inclusion of the separate sub-
tribes of Anatini defined by Livezey (1991) in the analysis was necessitated by
variation among the subgroups in characters determined to be informative as well
among “tadornine” genera. Under the taxonomy proposed by Livezey (1991),
the three subtribes of Anatini comprise the following genera: Cairineae {Cairina,
Pteronetta, andAix), Nettapodeae (Chenonetta and Nettapus), and Anateae (Ama-
zonetta, Callonetta, Lophonetta, Speculanas, Mareca, and Anas).
— —
34 Annals of Carnegie Museum VOL. 65
—
Composition ofIncluded Genera. -Five of the eight genera of primary interest
in this analysis are considered to comprise single species by most taxonomists:
Stictonetta, Plectropterus, Sarkidiornis, Hymenolaimus, and Merganetta (e.g., De-
lacour and Mayr, 1945; Delacour, 1954, 1956, 1959; Johnsgard, 1961a, 1965a,
\965b, 1978, 1979; Wolters, 1976). Segregation of Salvadori’s Duck in the genus
Salvadorina (Woolfenden, 1961; Kear, 1975; Wolters, 1976; Mlikovsky, 1989;
Sibley and Monroe, 1990; Livezey, 1991), instead of inclusion of the species
within Anas (e.g., Delacour and Mayr, 1945; Boetticher, 1952; Delacour, 1956;
Johnsgard, 1965a, 1978), delimits another monotypic genus in this group. —
Two of these purportedly monotypic genera^ Sarkidiornis and Merganetta
include two or more qualitatively distinct taxa that might be accorded species
rank. For example, in light of plumage differences and geographical isolation
(Madge and Burn, 1988), partitioning of Sarkidiornis into a nominate African
species S. melanotus and a Neotropical species S. sylvicola is consistent with both
the emphasis of the “minimal diagnosable unit” under the phylogenetic species
concept, and the requirements of distinct historical “identities” and independent
evolutionary “tendencies” and “fate” under the evolutionary species concept
(Wiley, 1978; Cracraft, 1983, 1988; McKitrick and Zink, 1988; Frost and Hillis,
1990). In practice, application of the phylogenetic species concept can approxi-
mate the lineages conceptualized under the evolutionary species concept (Frost
and Kluge, 1994). Numerous synapomorphies demonstrate the monophyly ofSar~
kidiornis (see beyond), however, and the included species were identical for the
characters coded here; therefore the species-level taxonomy of the genus is not
germane to this analysis. Similarly, substantial support for monophyly of Mer-
ganetta (see beyond) and identical states of analyzed characters render inconse-
quential any species-level interpretation one might impose on this poorly under-
stood complex of three to six recognized forms for purposes of this analysis
(Conover, 1943; Delacour, 1956; Johnsgard, 1978; Madge and Bum, 1988). The
steamer-ducks {Tachyeres) are treated as a single generic taxon for this analysis;
the genus comprises four confusingly similar species (Livezey and Humphrey,
1992), is demonstrably monophyletic (Livezey, 1986a, 1986Z?, prese'^nt study),
members of the genus were identical for included characters, and the phytogeny
of included taxa has been examined elsewhere (Livezey, 1986^?; Corbin et al.,
1988).
The Spur-winged Goose {Plectropterus gambensis) currently is considered to
comprise two poorly differentiated subspecies (northern, comparatively pale gam-
bensis and southern, comparatively dark niger) having a broad area of intergra-
dation; the two forms are considered as a single analytical taxon here (Delacour,
1959; Johnsgard, 1978; Brown et al., 1982; Madge and Burn, 1988). No diag-
nosable populations have been detected within the other four critical genera
Stictonetta, Hymenolaimus, Malacorhynchus, and Salvadorina (Delacour, 1956,
1959; Johnsgard, 1978, 1979).
Specimens and Related Data
—
Specimens Examined. Study skins of adult birds were compared for deter-
mination of plumage pattern and general structure of the externum in the collec-
tions of several major museums, most importantly the National Museum of Nat-
ural History and the American Museum of Natural History. Colors of the irides,
bill, tarsi, and feet of adults also were characterized using study skins, but colors
—
1996 Livezey Tadornine-Anatine Divergence 35
of unfeathered integument are not always reliably coded using dried specimens;
consequently, colors of soft parts were confirmed using published descriptions
(see below), notes on specimen labels, photographs of living individuals, and
direct observation ofliving birds in avicultural collections or in the wild. Plumage
patterns of downy young, specimens of which are comparatively rare in museum
collections, were ascertained by comparisons of study skins (including an impor-
tantcollection at the Wildfowl and Wetlands Trust) and fluid-preserved specimens.
Prepared skeletal specimens of adults were used for osteological comparisons.
Numbers of prepared skeletons available varied significantly among taxa. Large
series of skeletons (100 or more) were available for several groups (Tachyeres,
Anas), whereas fewer than five were available for several others {Hymenolaimus,
Merganetta); at least five skeletons were studied for other taxa, with the exception
of Salvadorina (for which no complete skeletal specimen exists). Tracheal char-
acters were determinedusing ossified elements in skeletal specimens and complete
tracheae removed from fluid-preserved specimens (some previously mounted and
dried). —
PublishedData. Published descriptions and illustrations of adult and subadult
plumages supplemented data collected from direct study (Conover, 1943; Dela-
cour, 1954, 1956, 1959, 1964; Johnsgard, 1965a, 1978; Weller, 1968; Todd, 1979;
Humphrey and Livezey, 1982; Madge and Burn, 1988; Marchant and Higgins,
1990; Hoyo et al., 1992). Preliminary study of variation in natal plumages was
accomplished using the illustrations and descriptions of downy young provided
by Delacour and Mayr (1945, 1946), Delacour (1954, 1956, 1959, 1964), Frith
(1955, 1964^7), Pengelly and Kear (1970), Kear (1972, 1975), and Humphrey and
Livezey (1985). Variation and alternative interpretations of skeletal characters,
most of which were based on my earlier analyses (Livezey, 1986a, 1989, 1991),
were made using the works by Shufeldt (1909), Verheyen (1953), Rand (1954),
Woolfenden (1961), Humphrey and Clark (1964), Cracraft (1968), Weller (1968),
Moller (1969), and Raikow (1970). The few skeletal characters determinable for
Salvadorina were coded using the descriptions of a few elements by Mayr (1931)
and Mlikovsky (1989). Additional information on the tracheal anatomy of water-
fowl was taken from Yarrell (1827), Garrod (1875), Campbell (1899), Mayr
(1931), Niethammer (1952), Johnsgard (1961^, 1971), and Humphrey and Clark
(1964).
Definition of Characters
—
Theory and Objectives. Characters used in this phylogenetic analysis are dis-
tinct morphological features, hypothesized to be homologous (de Pinna, 1991;
Lipscomb, 1992; Hall, 1994), that comprise two or more states within the ingroup
or differ between the ingroup and outgroups. Of the states defined, one is hy-
pothesized as the primitive or plesiomorphous state based on outgroup compari-
sons (see below), and the other(s) as the derived or apomorphic state(s). Definition
of characters and included states are essential to any cladistic analysis, but rec-
ognition of homology and description of the implicit distinctions between states
remain the most challenging and influential aspect of phylogenetic inference
(Pogue and Mickevich, 1990).
Given that shared derived characters (synapomorphies) constitute the only em-
pirical support for the delimitation ofhierarchically nested sets ofclosest relatives
(clades), the primary objective of character analysis is the discovery of such
36 Annals of Carnegie Museum VOL. 65
changes in states (transformation series) germane to the group of interest. Auta-
pomorphies, character changes that are unique to terminal taxa, cannot contribute
to the resolution of such groupings of included taxa (Wiley, 1981), although such
differences traditionally have been accorded considerable weight in previous at-
tempts at classification (e.g., Delacour, 1954; Johnsgard, 1961<3, 1965n, 1978).
However, autapomorphies ofthe key eight genera ofinterest were included in this
analysis for several reasons: (1) inclusion of autapomorphic divergence in cladis-
tically based phylograms can provide insights into relative evolutionary rates, (2)
autapomorphic characters ofpolytypic genera also are synapomorphic forincluded
members, and (3) autapomorphic divergence can be depicted in phylogenetic trees
independently of the character changes supporting topological structure and as-
sociated summary stati—stics.
A
Characters Defined. total of 114 morphological characters were defined for
this analysis: 39 characters of the skeleton, four of the trachea and syrinx, 11 of
the natal plumage, and 60 of the definitive plumage and soft parts (Appendix 1).
Most skeletal characters were adapted from those described by Livezey (1986(2),
with lesser numbers from Livezey (1989, 1991); for this analysis, states of these
characters were restricted to those represented by the ingroup. This more detailed
consideration of tracheosyringeal anatomy obviated the ad hoc weighting of the
single (composite) syringeal transformation series defined in the analysis by Liv-
ezey (1986(2).
Behavioral patterns, especially courtship displays, have received considerable
study in waterfowl (Delacour and Mayr, 1945; Delacour, 1954, 1956, 1959; Johns-
gard, 1960a, 1961(2, 1962, 1965a, 1978). Such ethological data have provided
insights into phylogenetic relationships of waterfowl (e.g., Johnsgard, 1961a) and
have been shown to be comparable in consistency to morphological and molecular
characters (de Queiroz and Wimberger, 1993). Formal inclusion of such data in
the phylogenetic analysis, however, was precluded by numerous missing data,
problems of coding behavioral characters termed “rare” or “uncommon” in a
discrete scheme, and questions of homology and polarity not addressable from
published descriptions.
Each character was coded for the 19 members ofthe ingroup and a hypothetical
“ancestor” (see below), producing a data matrix of dimension 20 X 114 (Ap-
pendix 2). Missing data were as follows: 28 entries for skeletal characters of
Salvadorina were coded as “missing”; two natal characters for the “hypothetical
ancestor” were coded as “missing,” indicating an unknown polarity; and one
entry for character 44, three entries for character 46, single entries for characters
47 and 48, and two entries for character 77 were treated as “missing” because
ofproblematic comparability ofstates. In total, there were 38 matrix entries coded
as “missing” (1.7% of matrix). The impact of missing data on phylogenetic in-
ference has been considered by Nixon and Davis (1991) and Platnick et al. (1991).
Fourteen characters were “multistate” (i.e., included two or more derived
states), of which two (characters 29 and 78) were ordered, i.e., a particular or-
dination of derived states was considered more probable on structural or devel-
opmental grounds (Wilkinson, 1992). Analytical implications of multistate char-
acters and ordering attract continuing controversy and empirical study (Hauser
and Presch, 1991; Mickevich and Lipscomb, 1991; Lipscomb, 1992; Barriel and
Tassy, 1993; Slowinski, 1993). Consequently, an assessment of the analytical im-
pact of these two orderings was made by comparison with analyses in which all
characters were treated as unordered.
